Fast Measurement of Blood T₁ in the Human Carotid Artery at 3 Tesla: Accuracy, Precision and Reproducibility

Abstract

Purpose

To develop a fast protocol for measuring T₁ values in the internal carotid artery (ICA), to validate this technique with in vitro measurements, and to evaluate its reproducibility.

Methods

A modified Look-Locker sequence was optimized to enable rapid determination of T₁ in the ICA at 3T. T₁ values from the ICA were compared with in vitro measurements on individually sampled venous blood oxygenated to arterial levels. A test-retest reproducibility study was also conducted.

Results

The group-averaged arterial blood T₁ value was 1908+/−77 ms for 6 female adults (hematocrit = 0.39+/−0.03) and 1785+/−55 ms for 7 male adults (hematocrit = 0.45+/−0.02), which is 100–200 ms longer than the widely adopted value obtained from bovine blood experiments. The arterial T₁ value per subject correlated significantly with individual hematocrit (Hct) values. The intra-session and inter-session coefficients of variation (CoV) were 1.1% and 2.1% respectively, indicating good precision and reproducibility of our method. Reasonable agreement was observed between the in vivo and in vitro results with a correlation coefficient of 0.78.

Conclusion

The proposed method can provide fast arterial T₁ measurement on individual subjects. When not performing such a subject-specific measurement, we recommend the use of 1908 ms and 1785 ms for healthy female and male adults, respectively, or 1841 ms for adults in general.

Introduction

The T₁ value of arterial blood is an important parameter for the quantification of cerebral blood flow (CBF) using arterial spin labeling (ASL) techniques (1). It is also critical in VASO experiments for relative cerebral blood volume (CBV) (2,3) or absolute arterial CBV (4,5) determination. Standard practice is to assume a constant arterial blood T₁ across all individuals (6). However, blood T₁ is known to have dependence on hematocrit (Hct) (7–19) (e.g. due to individual variation or in anemia or polycythemia), oxygenation fraction (Y) (9,11,13,16–25) (e.g. in hypoxemia or ischemia), the age of subject (7,8,16), and blood composition (26) (e.g. in sickle cell disease). Therefore, a fast protocol to measure individual arterial blood T₁ in vivo is desirable to accurately calculate CBF.

A modified Look-Locker scheme (7,8,12,15,27–29) has been proposed with a non-selective inversion pulse followed by a train of slice-selective readouts, through which multiple inversion time points can be acquired rapidly due to constant refreshing of blood. This method has been applied to measure Gd-dosed blood T₁ in renal arteries and veins (27,28) and venous blood T₁ in the sagittal sinus (8,15,29) and internal jugular vein (IJV) (12). However, the application of this approach to measure arterial blood T₁ suffers from the poor inversion efficiency for the arterial blood magnetization in the chest area, which is due to large B₀ and B₁ inhomogeneities and flow artifacts associated with the faster and more pulsatile flow of arterial blood.

In this work, we optimized our previous IJV-based T₁ protocol (12) to measure blood T₁ in the internal carotid artery (ICA), verified the accuracy of the results with in vitro experiments on individually sampled blood, and evaluated the precision and reproducibility of the new method with a test-retest study.

Methods

Fast T₁ Measurement in the ICA

The pulse sequence for measuring arterial blood T₁ at ICA was similar to the sequence used previously to measure venous blood T₁ at IJV (12), namely a nonselective adiabatic inversion pulse followed by multiple imaging acquisition blocks separated by short time intervals (T_d) (Fig. 1a). The main difference between a T₁ measurement at the ICA and IJV is the positioning of subjects relative to the RF body coil. The regular neuroimaging setup (centering at eye level) is sufficient for an IJV T₁ measurement, because the normal cerebral mean transit time (from ICA to IJV) is close to the TR ~ 10 s (30). Therefore all venous blood entering the IJV is within the brain when the inversion pulse is applied (12). For subjects without heart disease, the mean transit time from pulmonary artery to ICA is close to 8 sec (31,32). To acquire proper inversion recovery curves for the fast flowing arterial blood at the ICA, it is crucial to invert all of the incoming blood magnetization from the heart, lung, aorta, common carotid and ICA. Since the body coil length of the scanner used in this study was 65 cm, subjects needed to be positioned in the middle between ICA and diaphragm, i.e. the clavicle, to have all the feeding blood covered by the RF coil (Fig. 1b).

Figure 1.

a) Pulse diagram for measuring arterial T₁. Following an adiabatic 180° inversion pulse, a series TFE acquisitions are used to sample the inversion recovery curve with an interval that allows the blood magnetization to refresh in the imaging plane. THK=slice thickness; v=blood velocity. b) The survey image shows the B₁ coverage of the whole upper torso when centered at clavicle. c) Time of flight angiographic scout image. The red box indicates the imaging slice chosen to the ICA. d) one representative image from the fast T₁ protocol (TI=9850ms).

In addition, due to the larger B₀ and B₁ inhomogeneities in the chest area, the bandwidth of the standard hyperbolic secant (HS) adiabatic pulse (33) (10 ms, 1250 Hz bandwidth) used in the IJV T₁ protocol (12) did not achieve uniform inversion for the arterial T₁ measurement. HSn adiabatic pulses (34,35), which modify the HS pulse by raising the time variable in the sech function to a higher nth power leading to higher bandwidth, were investigated for better performance. Numerical simulations using the Bloch equations based on matrix rotation were conducted to assess the inversion efficiencies of different adiabatic pulses over a wide range of B₀/B₁ conditions: B₀ field: ±2000 Hz; B₁⁺ scale (ratio of actual flip angle to nominal input flip angle): from 0.0 to 2.0. Based on the simulation results, a 20 ms HS4 inversion pulse (n = 4, β = 4) was employed for in vivo experiments for its superior performance with 2500 Hz bandwidth through a large B₁⁺ scale range from 0.4 to 2.0 (Supporting Fig. S1).

In Vivo Experiments

Experiments were performed on a 3T Philips Achieva scanner (Philips Medical Systems, Best, The Netherlands) using the body coil for RF transmission (maximum B₁ amplitude 13.5 μT) and a 32-channel head coil for signal reception. Thirteen healthy volunteers (6 females, age: 35+/−9, 7 males, age: 35+/−6 years) participated in this study after providing informed written consent in accordance with the local Institutional Review Board guidelines.

As part of the scan planning, a survey image was acquired in the coronal plane to visualize the upper torso. A quick 3D Time-of-Flight (TOF) scan (0.5 min) was employed in the neck to visualize the location of neck vessels, including internal carotid arteries, external carotid arteries, and vertebral arteries. The field of view (FOV) was 180 × 160 × 99 mm³, the acquisition matrix was 180 × 159, and the reconstruction resolution was 0.94 × 0.93 × 3 mm³.

For the arterial blood T₁ measurements, an imaging plane of thickness (THK) = 5 mm was placed at the level of foramen magnum and perpendicular to one side of ICA above the bifurcation of the common carotid to reduce artifacts from the turbulence of arterial blood, as detailed in a previous study (36). The acquisition parameters were: FOV = 140 × 169 mm², acquisition matrix = 176 × 204, with left-right phase encoding. The in-plane resolution after reconstruction was 0.76 × 0.75 mm². Segmented turbo field echo (TFE) acquisition, instead of segmented EPI used previously (12), was chosen for its improved robustness against motion artifacts from arterial blood pulsation. The full k-space at each inversion time (TI) was acquired with 15 k-lines per segment (TFE factor = 15, low-high profile ordering), SENSE factor = 2, N_shot = 7. With flow compensation and TR/TE = 13/7.6 ms, the acquisition window (T_acq) was about 195 ms. The first TI was sampled at TI(1) = 160 ms and the others were acquired at intervals of 285 ms. Note that the gap between adjacent TFE acquisition blocks (T_d) should be longer than the time (THK/v) required for blood flowing through the slice with velocity v (Fig. 1a), in order to remove any spin history of previous excitations so that the inflowing spins are only affected by the inversion pulse (12). Using the previously measured velocity in the IJV (v = 16 cm/s) (12) as an estimate of the lower limit of the velocity at ICA, the threshold for the replenishing time for THK = 5 mm is 31 ms, which is much shorter than T_d = 90 ms. With a total of 34 TIs recorded, total TR per shot (TR_shot) was 9850 ms, leading to a total acquisition time of about 1 min.

Different excitation flip angles (10°, 30°, 50°, 70°, 90°) for the TFE acquisition were compared among 4 subjects. Higher flip angles suppressed the background static tissue better (Supporting Fig. S2a–e) as expected; the measured arterial T₁ values using these various flip angles were found consistent (Supporting Fig. S2f). Considering both signal-to-noise ratio (SNR) and specific absorption rate (SAR), 50° flip angle was chosen in the remaining experiments.

The reproducibility of this protocol was evaluated by scanning each volunteer in two sessions, one in the morning and another in the afternoon with a separation of 3–6 hours. During each session, T₁ measurements at ICA (subject centering at the clavicle) were each performed twice consecutively, using the acquisition protocol described above. Meanwhile, T₁ measurement at the IJV were also performed in every session based on previous protocol (12).

In Vitro Experiments

A subgroup of volunteers (6 females, 3 males, age: 35+/−8 years old) had their blood drawn right before or after their first MRI scans. Two 4mL tubes of venous blood were collected at the outpatient clinic of Johns Hopkins Hospital. One tube of blood (anticoagulated with K₂EDTA) was sent to the hospital’s pathology lab for a complete blood count (CBC). A second tube of blood (anticoagulated with lithium heparin) was brought back to our MRI center for the in vitro experiments on the same day, in order to validate the in vivo measurements. Before the MRI scans, the blood sample was oxygenated to 96%–99% through exposing it to the air typically for ~20 mins. After determining the oxygenation fraction (Y) and methemoglobin fraction (f_MetHb) concentration using the blood analyzer (ABL800, Radiometer), the sample was sealed in a tube with a diameter of 1 cm and equilibrated to 37.8°C.

A conventional inversion-recovery sequence (Supporting Fig. S3a) with TI = [0.2, 0.5, 1, 2, 3, 5, 9, 15] s was employed for the in vitro measurement of blood T₁. To ensure a constant initial magnetization before the inversion pulse for different TIs, a non-selective tailored hard-pulse saturation pulse train (65°, 83°, 143°, 162°) (37) following the acquisition reset the remaining magnetization to zero and a fixed delay of 2 s was inserted. Two-shot TFE (flip angle = 8°, TR/TE = 15/2.5 ms, TFE factor = 20, Tacq = 110 s) was used to acquire the 2D image of THK = 3 mm with the FOV of 64 × 120 mm² and resolution of 0.94 × 0.94 mm². The scan time was kept to within 2 min to prevent erythrocyte sedimentation. During the scan, the blood tube was placed in a 37.8 °C water bath (Supporting Fig. S3b,c) (16). For signal reception, a 2-channel shoulder coil was used and stabilized by foam pads. After the experiment, the temperature of the blood was measured (36.3–36.6 °C) and the oxygenation and lack of precipitation of erythrocytes were confirmed.

Data Analysis

Matlab (MathWorks, Inc., Natick, MA, USA) was used for simulation and data processing. ROIs were manually drawn on the images acquired at the longest TI and applied to images at other TIs.

It is important to point out that the slow flow near the vessel wall could make the fitted T₁ shorter if the blood in the imaging plane can not be refreshed before the next TFE acquisition block. Besides shortening T₁, such slow flow and partial volume effects with non-flowing water spins in surrounding tissue also reduce the imaging intensity for this thin layer. Therefore, for in vivo data, the center of the ICA vessels with higher signal were chosen as ROIs to avoid partial volume effect from the static tissue and the layer of blood with much slower velocity near the vessel wall. The averaged magnitude intensity of each ROI was fitted as a function of TI using the nonlinear-least-square algorithm for a 3-parameter model (12): the initial magnitude S₀, the inversion degree α and the blood T₁.

$$S(TI)=S_0|1-2\alpha e^{-TI/T_1}+e^{-T{R}_{\mathit{shot}}/T_1}|$$ [1]

For in vitro data with long TR, blood T₁ was fitted using a 2-parameter model (9), where C is related to saturation delay time and inversion efficiency:

$$S(TI)=S_0|1-C{e}^{-TI/T_1}|$$ [2]

In order to evaluate the reproducibility of our method, the coefficient of variation (CoV) for intra-session, inter-session and inter-subject were calculated as:

$$Co{V}_{\mathit{IntraSession}}=\frac{1}{I\times J}{\sum }_{i=1}^I{\sum }_{j=1}^J\frac{|T_1^{ij1}-T_1^{ij2}|}{\sqrt{2}·\mathit{Mean}(T_1^{ij1},T_1^{ij2})}$$ [3]

$$Co{V}_{\mathit{InterSession}}=\frac{1}{I\times K}{\sum }_{i=1}^I{\sum }_{k=1}^K\frac{|T_1^{i1k}-T_1^{i2k}|}{\sqrt{2}·\mathit{Mean}(T_1^{i1k},T_1^{i2k})}$$ [4]

$$Co{V}_{\mathit{InterSubject}}=\frac{1}{J\times K}{\sum }_{j=1}^J{\sum }_{k=1}^K\frac{ST{D}_i(T_1^{\mathit{ijk}})}{\mathit{Mean}(T_1^{1jk},T_1^{2jk})}$$ [5]

where T₁^ijk represents the kth measurement of blood T₁ value at participant #i in session #j. I, J, and K were total number of participants, sessions and measurement times in one session, respectively. STD_i(T₁^ijk) stands for the standard deviation of T₁ values for all the subjects at the kth measurement in session #j.

Results

Fig. 1b–d show typical localization of the imaging slice for the T₁ measurement at ICA, prescribed on both the anatomical (Fig. 1b) and angiographic images (Fig. 1c). When subjects were centered at the clavicle, reasonable inversion recovery data (Supporting Fig. S4) and arterial blood T₁ values were obtained for the ICA (Fig. 2a, red curve). However, when subjects were centered at the eye level, the inflow of non-inverted fresh blood reduced the measured T₁ at ICA notably (Fig. 2a, black solid curve). Fig. 2b compares results using a 10 ms HS pulse with bandwidth = 1250 Hz (black dash curve) and a 20 ms HS4 pulse with bandwidth = 2500 Hz (red solid curve). The HS4 pulse with much larger B₀/B₁ tolerance (Supporting Fig. S1) yielded more acceptable T₁ values from ICA blood (Fig. 2b). It is worth noting that any metal in necklaces and the bra (breast retainment apparatus), which did not cause problems for T₁ measurements at IJV, were found to hugely affect the B₀ inhomogeneity in the chest area and thus demand removal before scans. No motion artifacts were observed for any data sets.

Figure 2.

a) Inversion recovery curves measured at the same subject with the same pulse sequence but with the subject positioned at the eye (black line) and at the clavicle (red line). b) Inversion recovery curves measured using HS and HS4 inversion pulses with the same subject positioned at the clavicle.

Table 1 reports the hematocrits (Hct) and hemoglobin concentrations in whole blood (ctHb) from the CBC tests, along with the arterial and venous T₁ values of each subject averaged from the four measurements in the morning and afternoon sessions. The averaged arterial T₁ values were 1908+/−77 ms for female participants (n = 6, Hct = 0.39+/−0.03, ctHb = 13.2±1.0 g/dL), and 1785+/−55 ms for male participants (n = 7, Hct = 0.45+/−0.02, ctHb = 15.1±1.0 g/dL). For all the participants combined (n = 13, Hct = 0.42+/−0.04, ctHb = 14.2±1.4 g/dL), the averaged arterial T₁ value of 1841+/−89 ms was 56+/−35 ms longer than the average venous T₁ value of 1785+/−80 ms. The in vitro arterial T₁ values from 9 subjects are also listed in Table 1.

Table 1.

The arterial T₁ measured from internal carotid artery and in vitro for different subjects.

Subject		Hct	ctHb (g/dL)	T1a,in vivo (ms)	T1v,in vivo (ms)	fMetHb	T1a,in vitro (ms)
1	Female	0.40	13.4	1876±19	1835±32	1.0%	1851±19
2	Female	0.40	13.5	1921±41	1811±37	0.9%	1825±41
3	Female	0.36	11.7	2037±26	1966±28	1.0%	1990±26
4	Female	0.44	14.8	1854±28	1760±35	0.9%	1927±28
5	Female	0.38	12.9	1820±30	1749±27	1.0%	1910±30
6	Female	0.39	12.8	1937±53	1862±70	1.1%	1906±53
7	Male	0.47	15.9	1720±26	1685±34	0.6%	1730±26
8	Male	0.45	15.6	1804±61	1771±17	1.1%	1806±61
9	Male	0.44	14.6	1811±30	1727±37	1.0%	1797±30
10	Male	0.47	16.3	1725±41	1672±15
11	Male	0.41	13.4	1880±17	1846±18
12	Male	0.46	15.7	1789±24	1730±8
13	Male	0.43	14.4	1766±21	1792±13
Female Mean±SD		0.39±0.03	13.2±1.0	1908±77	1831±79	1.0±0.1%	1902±58
Male Mean±SD		0.45±0.02	15.1±1.0	1785±55	1746±61	0.9±0.3%	1778±41
Total Mean±SD		0.42±0.04	14.2±1.4	1841±89	1785±80	1.0±0.2%	1860±80

As shown in the scatter plot (Fig. 3a) and Bland-Altman (Fig. 3b) plots, reasonable correlation and agreement between arterial T₁ values measured in vivo and in vitro were found with a correlation coefficient of 0.78. Additionally, the in vivo arterial T₁ values were found to differ significantly (P = 6.1×10⁻⁵) from the in vivo venous T₁ values (Fig. 3c). As expected, the arterial T₁ values measured in vivo correlated significantly with Hct (Fig. 4a and Eq. [6]) as well as with ctHb (Fig. 4b and Eq. [7]).

Figure 3.

a) correlation and b) agreement (Bland-Altman plot) between arterial blood T₁ values measured in vivo and in vitro. In a), the solid line is a linear fit of the arterial blood T₁ values measured in vivo vs in vitro. In b), the solid line indicates the mean difference between arterial blood T₁ values measured in vivo vs in vitro at the same subject, and the dashed lines represent two times the standard deviation from the mean difference. c) Bland-Altman plot to compare arterial and venous T₁ measured in vivo. The solid line indicates the mean difference between arterial blood T₁ and venous blood T₁, and the dashed lines represent two times the standard deviation from the mean difference.

Figure 4.

Linear relationship between the arterial longitudinal relaxation rate (R_1a = 1/T_1a) and Hct (a) and ctHb (b).

$$1000/{\mathrm{T}}_{1\mathrm{a}}(\text{ms})=0.625\times \text{Hct}+0.280$$ [6]

$$1000/{\mathrm{T}}_{1\mathrm{a}}(\text{ms})=0.016\times \text{ctHb}(\mathrm{g}/\text{dL})+0.317$$ [7]

In the reproducibility study, the intra-session CoV was 1.1%, which reflects the low measurement noise and high reliability of this technique. The inter-session CoV was 2.1%, which is a little higher than intra-session CoV. We tentatively attribute this to possible physiological fluctuation (e.g. fluid intake) between morning and afternoon. The inter-subject CoV was 5.6%, which reflects the physiological differences (e.g. in Hct) between different subjects.

Discussion

A new approach for rapid measurement of T₁ values from the ICA at 3T was presented through optimization of body positioning, inversion pulse type and acquisition scheme of a previous protocol to measure blood T₁ in the IJV (12). Specifically, the subjects were centered at the clavicle, and an inversion pulse (HS4) with reduced sensitivity to B₀/B₁ inhomogeneity was employed to ensure effective inversion of the magnetization of feeding arterial blood. Furthermore, TFE acquisition with large flip angles was used to suppress the background signal while maintaining high SNR. The accuracy of our in vivo method was validated by comparing the measured T₁ values with in vitro data. Low intra-session and inter-session CoVs, 1.1% and 2.1% respectively, indicated excellent precision and reproducibility.

The measured arterial blood T₁ values of males (1785+/−55 ms, Hct = 0.45+/−0.02) agree well with a previous arterial T₁ study on the abdominal aorta of male subjects (1779+/−80 ms with Hct = 0.47+/−0.03) using a traditional inversion-recovery method (11). For both females and males, the measured arterial blood T₁ values (1841+/−89 ms, Hct = 0.42+/−0.04, Table 1) are close to the arterial value predicted from our recent theoretical T₁ model (1839 ms, Supporting Table S1) (19) and another model calibrated from in vivo data (1771 ms) (38) for normal blood (Hct = 0.42, Y = 100%, and f_MetHb = 1.0%,), but are about 200 ms longer than the widely adopted value (1664 ms with Hct = 0.42) based on bovine blood experiments (13). This ~200 ms difference is partly due to a higher level of MetHb in bovine samples. The bovine blood had about ~2% MetHb vs. 1.0+/−0.2% for the human blood sampled in this study (Table 1). According to the T₁ model (19), an increase of 1% in MetHb induces ~100 ms shortening in blood T₁ (Supporting Table S1).

The blood T₁ values measured at the ICA were 56+/−35 ms longer than the ones at the IJV (Table 1, Fig. 3c). Interestingly, when using the T₁ model (19), the T₁ values of arterial (Y: 100%) and venous (Y: 65%) blood were predicted as 1839 ms and 1681 ms, respectively, with a 158 ms difference between them (Supporting Table S1). One possible explanation for the smaller difference of arterial-venous T₁ values observed in this study may be related to the fact that some portion of the blood contributing to the later part of the inversion recovery curve originated from the capillary compartment. Note that the microvascular Hct in capillaries is known to be approximately 75%–85% of that for large vessels (Fahraeus effect) (39,40). Based on our T₁ model (19), a lower Hct level (e.g., Hct = 0.80 × 0.42 = 0.34) with increased oxygenation fraction (e.g., Y = (100% + 65%)/2 = 82%) gives an increased blood T₁, yielding a capillary T₁ of 1857 ms (Supporting Table S1), which is close to our theoretical arterial T₁ (1839 ms, Supporting Table S1). Accordingly, this contamination of blood from capillaries may augment the T₁ estimated at IJV but negligibly impact the T₁ measurement at ICA. Although shortening the repetition time of the employed inversion recovery sequence would be desirable to reduce this bias, the data acquired at longer TIs are close to the equilibrium magnetization and could allow T₁ fitting more robust against physiology noise.

Other possible causes of the lower arterial T₁ may be incomplete refreshing of spins or inhomogeneous inversion for the chest. We think we can exclude the first based on arguments in the methods section for spin history, but incomplete inversion coverage of the thoracic cavity where the blood supply for the ICA is located may be possible. However, the arterial T₁ values agree well with in vitro data (Fig. 3a,b, Table 1), which suggests that this was not a significant source of error in this study. Admittedly, for regular neuroimaging studies, the described method for obtaining arterial T₁ values by repositioning to the subject’s clavicle is not as efficient and convenient as the protocols for measuring venous T₁ (8,12,15). Therefore, when scanning a person with normal blood composition, one alternative could be to first measure the T₁ at IJV (12), and then empirically add 56 ms found in this study to estimate the arterial T₁ value.

The significant difference of arterial blood T₁ values found between female and male adults (1908±77 ms vs. 1785+/−55 ms) was shown to be associated with the gender difference of their Hct values (0.39+/−0.03 vs. 0.45+/−0.02) (Table 1, Fig. 4a). As demonstrated in previous studies (7–19) on both arterial and venous blood, the blood relaxation rate 1/T₁ has a linear dependence on Hct. Earlier work (10,19,20) has asserted hemoglobin as the main source of the T₁ relaxation in arterial blood. Indeed, the good correlation of 1/T₁ and Hct is due to the narrow distribution of the mean corpuscular hemoglobin concentration (MCHC), which is the concentration of hemoglobin in erythrocytes and ranges from 32~36 g/dL in normal blood (41). Therefore, the correlation (Fig 4b) between the 1/T₁ and ctHb (ctHb = Hct · MCHC) is more instructive, especially for the abnormal blood such as sickle cell disease where erythrocytes are severely dehydrated and MCHC is elevated (42).

One question is whether pulsatile flow may still significantly affect the measured T₁. We do not think so because during the interval (T_d in Figure 1) between adjacent TFE acquisitions, all blood in the imaging plane will be refreshed even at the diastolic rate. Future cardiac-gated measurements could provide more insight into this. However, pulsatile flow may affect image quality because we only had first order flow compensation during the acquisition. Indeed, we saw improvements in the images when replacing the segmented EPI acquisitions used in our previous IJV T₁ paper to the current segmented TFE acquisitions.

Knowledge of the arterial T₁ is critical for quantitative CBF mapping with ASL (6,43,44) and CBV studies with VASO (4,5,45). It is known that CBF values estimated using ASL are negatively correlated with the arterial blood T₁ values (6). Currently, the arterial blood T₁ of the healthy adults to be used for the quantification of ASL data is recommended as 1650 ms (13) at 3.0T. For the highest arterial blood T₁ value obtained in this study (2037 ms, Hct: 0.36), using this 18% undervalued blood T₁ (1650 ms) would cause a 35% overestimation of CBF. Similarly, the inversion time in VASO experiment will be underestimated by 18% and thus cause a 70% overestimation of ΔCBV.

Conclusion

We demonstrated a fast technique (~1 min) for accurately and reliably measuring arterial blood T₁ at the ICA. The proposed method can be performed individually to avoid possible inaccuracies due to Hct variation or abnormal blood composition. When not performing such a measurement, we recommend the use of 1908 ms and 1785 ms for healthy female and male adults or 1841 ms for adults in general, or estimating the arterial blood T₁ values based on Eqs. [6] or [7] if Hct or ctHb are known for subjects with normal blood composition. The combination of our method with ASL and VASO techniques can result in more accurate estimation of CBF and CBV, especially for the patients with blood disorders.

Supplementary Material

Supp Info

Supporting Figure S1. a) The amplitude and frequency modulation of the respective HS (10 ms, bandwidth: 1250 Hz) and HS4 inversion pulse (20 ms, n = 4, β = 4, bandwidth: 2500 Hz), and b) the simulated M_z responses to different B₀ off-resonance frequencies and B₁⁺ scales for each pulse. The HS4 pulse was chosen for the in vivo experiments due to its broader bandwidth and adequate adiabaticity.

Supporting Figure S2 Representative images from the fast T₁ protocol (TI = 9850 ms) for different excitation flip angles (a–e) and the measured T₁ values with different flip angles from four subjects (different colors represents different subjects).

Supporting Figure S3. The pulse sequence (a), experimental setup (b), a representative imaging slice (c) and a representative inversion recovery curve (d) for the T₁ measurement of in vitro blood. The blood sample was contained in a 1 cm-diameter tube, which was fixed in the black lid of the white jar. The white jar was filled with 37 °C water to keep blood sample temperature during the experiment. Three bottles with Copper sulfate solution were used to load the coil.

Supporting Figure S4. Images in one in vivo arterial T₁ inversion recovery measurement from a representative subject. TI of 34 images were continuously increasing and evenly sampled with the first image’s TI was 160ms and the last image’s TI was 9582ms.

Supporting Table S1. The blood T₁ predicted based on the T₁ model (19).